languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21 use MediaWiki\MediaWikiServices;
  22
  23 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         static public $languagesWithVariants = [
  41                 'en',
  42                 'crh',
  43                 'gan',
  44                 'iu',
  45                 'kk',
  46                 'ku',
  47                 'shi',
  48                 'sr',
  49                 'tg',
  50                 'uz',
  51                 'zh',
  52         ];
  53
  54         public $mMainLanguageCode;
  55
  56         /**
  57          * @var string[]
  58          */
  59         public $mVariants;
  60         public $mVariantFallbacks;
  61         public $mVariantNames;
  62         public $mTablesLoaded = false;
  63         public $mTables;
  64         // 'bidirectional' 'unidirectional' 'disable' for each variant
  65         public $mManualLevel;
  66
  67         public $mLangObj;
  68         public $mFlags;
  69         public $mDescCodeSep = ':', $mDescVarSep = ';';
  70         public $mUcfirst = false;
  71         public $mConvRuleTitle = false;
  72         public $mURLVariant;
  73         public $mUserVariant;
  74         public $mHeaderVariant;
  75         public $mMaxDepth = 10;
  76         public $mVarSeparatorPattern;
  77
  78         const CACHE_VERSION_KEY = 'VERSION 7';
  79
  80         /**
  81          * @param Language $langobj
  82          * @param string $maincode The main language code of this language
  83          * @param string[] $variants The supported variants of this language
  84          * @param array $variantfallbacks The fallback language of each variant
  85          * @param array $flags Defining the custom strings that maps to the flags
  86          * @param array $manualLevel Limit for supported variants
  87          */
  88         public function __construct( Language $langobj, $maincode, $variants = [],
  89                                                                 $variantfallbacks = [], $flags = [],
  90                                                                 $manualLevel = [] ) {
  91                 global $wgDisabledVariants;
  92                 $this->mLangObj = $langobj;
  93                 $this->mMainLanguageCode = $maincode;
  94                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  95                 $this->mVariantFallbacks = $variantfallbacks;
  96                 $this->mVariantNames = Language::fetchLanguageNames();
  97                 $defaultflags = [
  98                         // 'S' show converted text
  99                         // '+' add rules for alltext
 100                         // 'E' the gave flags is error
 101                         // these flags above are reserved for program
 102                         'A' => 'A',   // add rule for convert code (all text convert)
 103                         'T' => 'T',   // title convert
 104                         'R' => 'R',   // raw content
 105                         'D' => 'D',   // convert description (subclass implement)
 106                         '-' => '-',   // remove convert (not implement)
 107                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 108                         'N' => 'N',   // current variant name
 109                 ];
 110                 $this->mFlags = array_merge( $defaultflags, $flags );
 111                 foreach ( $this->mVariants as $v ) {
 112                         if ( array_key_exists( $v, $manualLevel ) ) {
 113                                 $this->mManualLevel[$v] = $manualLevel[$v];
 114                         } else {
 115                                 $this->mManualLevel[$v] = 'bidirectional';
 116                         }
 117                         $this->mFlags[$v] = $v;
 118                 }
 119         }
 120
 121         /**
 122          * Get all valid variants.
 123          * Call this instead of using $this->mVariants directly.
 124          *
 125          * @return string[] Contains all valid variants
 126          */
 127         public function getVariants() {
 128                 return $this->mVariants;
 129         }
 130
 131         /**
 132          * In case some variant is not defined in the markup, we need
 133          * to have some fallback. For example, in zh, normally people
 134          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 135          * when zh-sg is preferred but not defined, we will pick zh-hans
 136          * in this case. Right now this is only used by zh.
 137          *
 138          * @param string $variant The language code of the variant
 139          * @return string|array The code of the fallback language or the
 140          *   main code if there is no fallback
 141          */
 142         public function getVariantFallbacks( $variant ) {
 143                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 144                         return $this->mVariantFallbacks[$variant];
 145                 }
 146                 return $this->mMainLanguageCode;
 147         }
 148
 149         /**
 150          * Get the title produced by the conversion rule.
 151          * @return string The converted title text
 152          */
 153         public function getConvRuleTitle() {
 154                 return $this->mConvRuleTitle;
 155         }
 156
 157         /**
 158          * Get preferred language variant.
 159          * @return string The preferred language code
 160          */
 161         public function getPreferredVariant() {
 162                 global $wgDefaultLanguageVariant, $wgUser;
 163
 164                 $req = $this->getURLVariant();
 165
 166                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 167
 168                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 169                         $req = $this->getUserVariant();
 170                 } elseif ( !$req ) {
 171                         $req = $this->getHeaderVariant();
 172                 }
 173
 174                 if ( $wgDefaultLanguageVariant && !$req ) {
 175                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 176                 }
 177
 178                 $req = $this->validateVariant( $req );
 179
 180                 // This function, unlike the other get*Variant functions, is
 181                 // not memoized (i.e. there return value is not cached) since
 182                 // new information might appear during processing after this
 183                 // is first called.
 184                 if ( $req ) {
 185                         return $req;
 186                 }
 187                 return $this->mMainLanguageCode;
 188         }
 189
 190         /**
 191          * Get default variant.
 192          * This function would not be affected by user's settings
 193          * @return string The default variant code
 194          */
 195         public function getDefaultVariant() {
 196                 global $wgDefaultLanguageVariant;
 197
 198                 $req = $this->getURLVariant();
 199
 200                 if ( !$req ) {
 201                         $req = $this->getHeaderVariant();
 202                 }
 203
 204                 if ( $wgDefaultLanguageVariant && !$req ) {
 205                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 206                 }
 207
 208                 if ( $req ) {
 209                         return $req;
 210                 }
 211                 return $this->mMainLanguageCode;
 212         }
 213
 214         /**
 215          * Validate the variant
 216          * @param string|null $variant The variant to validate
 217          * @return mixed Returns the variant if it is valid, null otherwise
 218          */
 219         public function validateVariant( $variant = null ) {
 220                 if ( $variant === null ) {
 221                         return null;
 222                 }
 223                 // Our internal variants are always lower-case; the variant we
 224                 // are validating may have mixed case.
 225                 $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) );
 226                 if ( in_array( $variant, $this->mVariants ) ) {
 227                         return $variant;
 228                 }
 229                 // Browsers are supposed to use BCP 47 standard in the
 230                 // Accept-Language header, but not all of our internal
 231                 // mediawiki variant codes are BCP 47.  Map BCP 47 code
 232                 // to our internal code.
 233                 foreach ( $this->mVariants as $v ) {
 234                         // Case-insensitive match (BCP 47 is mixed case)
 235                         if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) {
 236                                 return $v;
 237                         }
 238                 }
 239                 return null;
 240         }
 241
 242         /**
 243          * Get the variant specified in the URL
 244          *
 245          * @return mixed Variant if one found, null otherwise
 246          */
 247         public function getURLVariant() {
 248                 global $wgRequest;
 249
 250                 if ( $this->mURLVariant ) {
 251                         return $this->mURLVariant;
 252                 }
 253
 254                 // see if the preference is set in the request
 255                 $ret = $wgRequest->getText( 'variant' );
 256
 257                 if ( !$ret ) {
 258                         $ret = $wgRequest->getVal( 'uselang' );
 259                 }
 260
 261                 $this->mURLVariant = $this->validateVariant( $ret );
 262                 return $this->mURLVariant;
 263         }
 264
 265         /**
 266          * Determine if the user has a variant set.
 267          *
 268          * @return mixed Variant if one found, null otherwise
 269          */
 270         protected function getUserVariant() {
 271                 global $wgUser;
 272
 273                 // memoizing this function wreaks havoc on parserTest.php
 274                 /*
 275                 if ( $this->mUserVariant ) {
 276                         return $this->mUserVariant;
 277                 }
 278                 */
 279
 280                 // Get language variant preference from logged in users
 281                 // Don't call this on stub objects because that causes infinite
 282                 // recursion during initialisation
 283                 if ( !$wgUser->isSafeToLoad() ) {
 284                         return false;
 285                 }
 286                 if ( $wgUser->isLoggedIn() ) {
 287                         if (
 288                                 $this->mMainLanguageCode ==
 289                                 MediaWikiServices::getInstance()->getContentLanguage()->getCode()
 290                         ) {
 291                                 $ret = $wgUser->getOption( 'variant' );
 292                         } else {
 293                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 294                         }
 295                 } else {
 296                         // figure out user lang without constructing wgLang to avoid
 297                         // infinite recursion
 298                         $ret = $wgUser->getOption( 'language' );
 299                 }
 300
 301                 $this->mUserVariant = $this->validateVariant( $ret );
 302                 return $this->mUserVariant;
 303         }
 304
 305         /**
 306          * Determine the language variant from the Accept-Language header.
 307          *
 308          * @return mixed Variant if one found, null otherwise
 309          */
 310         protected function getHeaderVariant() {
 311                 global $wgRequest;
 312
 313                 if ( $this->mHeaderVariant ) {
 314                         return $this->mHeaderVariant;
 315                 }
 316
 317                 // See if some supported language variant is set in the
 318                 // HTTP header.
 319                 $languages = array_keys( $wgRequest->getAcceptLang() );
 320                 if ( empty( $languages ) ) {
 321                         return null;
 322                 }
 323
 324                 $fallbackLanguages = [];
 325                 foreach ( $languages as $language ) {
 326                         $this->mHeaderVariant = $this->validateVariant( $language );
 327                         if ( $this->mHeaderVariant ) {
 328                                 break;
 329                         }
 330
 331                         // To see if there are fallbacks of current language.
 332                         // We record these fallback variants, and process
 333                         // them later.
 334                         $fallbacks = $this->getVariantFallbacks( $language );
 335                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 336                                 $fallbackLanguages[] = $fallbacks;
 337                         } elseif ( is_array( $fallbacks ) ) {
 338                                 $fallbackLanguages =
 339                                         array_merge( $fallbackLanguages, $fallbacks );
 340                         }
 341                 }
 342
 343                 if ( !$this->mHeaderVariant ) {
 344                         // process fallback languages now
 345                         $fallback_languages = array_unique( $fallbackLanguages );
 346                         foreach ( $fallback_languages as $language ) {
 347                                 $this->mHeaderVariant = $this->validateVariant( $language );
 348                                 if ( $this->mHeaderVariant ) {
 349                                         break;
 350                                 }
 351                         }
 352                 }
 353
 354                 return $this->mHeaderVariant;
 355         }
 356
 357         /**
 358          * Dictionary-based conversion.
 359          * This function would not parse the conversion rules.
 360          * If you want to parse rules, try to use convert() or
 361          * convertTo().
 362          *
 363          * @param string $text The text to be converted
 364          * @param bool|string $toVariant The target language code
 365          * @return string The converted text
 366          */
 367         public function autoConvert( $text, $toVariant = false ) {
 368                 $this->loadTables();
 369
 370                 if ( !$toVariant ) {
 371                         $toVariant = $this->getPreferredVariant();
 372                         if ( !$toVariant ) {
 373                                 return $text;
 374                         }
 375                 }
 376
 377                 if ( $this->guessVariant( $text, $toVariant ) ) {
 378                         return $text;
 379                 }
 380                 /* we convert everything except:
 381                    1. HTML markups (anything between < and >)
 382                    2. HTML entities
 383                    3. placeholders created by the parser
 384                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 385                    Minimize use of backtracking where possible.
 386                 */
 387                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 388
 389                 // this one is needed when the text is inside an HTML markup
 390                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 391
 392                 // Optimize for the common case where these tags have
 393                 // few or no children. Thus try and possesively get as much as
 394                 // possible, and only engage in backtracking when we hit a '<'.
 395
 396                 // disable convert to variants between <code> tags
 397                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 398                 // disable conversion of <script> tags
 399                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 400                 // disable conversion of <pre> tags
 401                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 402                 // The "|.*+)" at the end, is in case we missed some part of html syntax,
 403                 // we will fail securely (hopefully) by matching the rest of the string.
 404                 $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 405
 406                 $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 407                         '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 408                 $startPos = 0;
 409                 $sourceBlob = '';
 410                 $literalBlob = '';
 411
 412                 // Guard against delimiter nulls in the input
 413                 // (should never happen: see T159174)
 414                 $text = str_replace( "\000", '', $text );
 415                 $text = str_replace( "\004", '', $text );
 416
 417                 $markupMatches = null;
 418                 $elementMatches = null;
 419
 420                 // We add a marker (\004) at the end of text, to ensure we always match the
 421                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 422                 while ( $startPos < strlen( $text ) ) {
 423                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 424                                 $elementPos = $markupMatches[0][1];
 425                                 $element = $markupMatches[0][0];
 426                                 if ( $element === "\004" ) {
 427                                         // We hit the end.
 428                                         $elementPos = strlen( $text );
 429                                         $element = '';
 430                                 } elseif ( substr( $element, -1 ) === "\004" ) {
 431                                         // This can sometimes happen if we have
 432                                         // unclosed html tags (For example
 433                                         // when converting a title attribute
 434                                         // during a recursive call that contains
 435                                         // a &lt; e.g. <div title="&lt;">.
 436                                         $element = substr( $element, 0, -1 );
 437                                 }
 438                         } else {
 439                                 // If we hit here, then Language Converter could be tricked
 440                                 // into doing an XSS, so we refuse to translate.
 441                                 // If non-crazy input manages to reach this code path,
 442                                 // we should consider it a bug.
 443                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 444                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 445                                         . ". Disabling language conversion for this page.",
 446                                         [
 447                                                 "method" => __METHOD__,
 448                                                 "variant" => $toVariant,
 449                                                 "startOfText" => substr( $text, 0, 500 )
 450                                         ]
 451                                 );
 452                                 return $text;
 453                         }
 454                         // Queue the part before the markup for translation in a batch
 455                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 456
 457                         // Advance to the next position
 458                         $startPos = $elementPos + strlen( $element );
 459
 460                         // Translate any alt or title attributes inside the matched element
 461                         if ( $element !== ''
 462                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 463                         ) {
 464                                 // FIXME, this decodes entities, so if you have something
 465                                 // like <div title="foo&lt;bar"> the bar won't get
 466                                 // translated since after entity decoding it looks like
 467                                 // unclosed html and we call this method recursively
 468                                 // on attributes.
 469                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 470                                 // Ensure self-closing tags stay self-closing.
 471                                 $close = substr( $elementMatches[2], -1 ) === '/' ? ' /' : '';
 472                                 $changed = false;
 473                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 474                                         if ( !isset( $attrs[$attrName] ) ) {
 475                                                 continue;
 476                                         }
 477                                         $attr = $attrs[$attrName];
 478                                         // Don't convert URLs
 479                                         if ( !strpos( $attr, '://' ) ) {
 480                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 481                                         }
 482
 483                                         if ( $attr !== $attrs[$attrName] ) {
 484                                                 $attrs[$attrName] = $attr;
 485                                                 $changed = true;
 486                                         }
 487                                 }
 488                                 if ( $changed ) {
 489                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 490                                                 $close . $elementMatches[3];
 491                                 }
 492                         }
 493                         $literalBlob .= $element . "\000";
 494                 }
 495
 496                 // Do the main translation batch
 497                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 498
 499                 // Put the output back together
 500                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 501                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 502                 $output = '';
 503                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 504                         $output .= $translatedIter->current();
 505                         $output .= $literalIter->current();
 506                         $translatedIter->next();
 507                         $literalIter->next();
 508                 }
 509
 510                 return $output;
 511         }
 512
 513         /**
 514          * Translate a string to a variant.
 515          * Doesn't parse rules or do any of that other stuff, for that use
 516          * convert() or convertTo().
 517          *
 518          * @param string $text Text to convert
 519          * @param string $variant Variant language code
 520          * @return string Translated text
 521          */
 522         public function translate( $text, $variant ) {
 523                 // If $text is empty or only includes spaces, do nothing
 524                 // Otherwise translate it
 525                 if ( trim( $text ) ) {
 526                         $this->loadTables();
 527                         $text = $this->mTables[$variant]->replace( $text );
 528                 }
 529                 return $text;
 530         }
 531
 532         /**
 533          * Call translate() to convert text to all valid variants.
 534          *
 535          * @param string $text The text to be converted
 536          * @return array Variant => converted text
 537          */
 538         public function autoConvertToAllVariants( $text ) {
 539                 $this->loadTables();
 540
 541                 $ret = [];
 542                 foreach ( $this->mVariants as $variant ) {
 543                         $ret[$variant] = $this->translate( $text, $variant );
 544                 }
 545
 546                 return $ret;
 547         }
 548
 549         /**
 550          * Apply manual conversion rules.
 551          *
 552          * @param ConverterRule $convRule
 553          */
 554         protected function applyManualConv( $convRule ) {
 555                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 556                 // title conversion.
 557                 // T26072: $mConvRuleTitle was overwritten by other manual
 558                 // rule(s) not for title, this breaks the title conversion.
 559                 $newConvRuleTitle = $convRule->getTitle();
 560                 if ( $newConvRuleTitle ) {
 561                         // So I add an empty check for getTitle()
 562                         $this->mConvRuleTitle = $newConvRuleTitle;
 563                 }
 564
 565                 // merge/remove manual conversion rules to/from global table
 566                 $convTable = $convRule->getConvTable();
 567                 $action = $convRule->getRulesAction();
 568                 foreach ( $convTable as $variant => $pair ) {
 569                         $v = $this->validateVariant( $variant );
 570                         if ( !$v ) {
 571                                 continue;
 572                         }
 573
 574                         if ( $action == 'add' ) {
 575                                 // More efficient than array_merge(), about 2.5 times.
 576                                 foreach ( $pair as $from => $to ) {
 577                                         $this->mTables[$v]->setPair( $from, $to );
 578                                 }
 579                         } elseif ( $action == 'remove' ) {
 580                                 $this->mTables[$v]->removeArray( $pair );
 581                         }
 582                 }
 583         }
 584
 585         /**
 586          * Auto convert a Title object to a readable string in the
 587          * preferred variant.
 588          *
 589          * @param Title $title A object of Title
 590          * @return string Converted title text
 591          */
 592         public function convertTitle( $title ) {
 593                 $variant = $this->getPreferredVariant();
 594                 $index = $title->getNamespace();
 595                 if ( $index !== NS_MAIN ) {
 596                         $text = $this->convertNamespace( $index, $variant ) . ':';
 597                 } else {
 598                         $text = '';
 599                 }
 600                 $text .= $this->translate( $title->getText(), $variant );
 601                 return $text;
 602         }
 603
 604         /**
 605          * Get the namespace display name in the preferred variant.
 606          *
 607          * @param int $index Namespace id
 608          * @param string|null $variant Variant code or null for preferred variant
 609          * @return string Namespace name for display
 610          */
 611         public function convertNamespace( $index, $variant = null ) {
 612                 if ( $index === NS_MAIN ) {
 613                         return '';
 614                 }
 615
 616                 if ( $variant === null ) {
 617                         $variant = $this->getPreferredVariant();
 618                 }
 619
 620                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 621                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 622                 $nsVariantText = $cache->get( $key );
 623                 if ( $nsVariantText !== false ) {
 624                         return $nsVariantText;
 625                 }
 626
 627                 // First check if a message gives a converted name in the target variant.
 628                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 629                 if ( $nsConvMsg->exists() ) {
 630                         $nsVariantText = $nsConvMsg->plain();
 631                 }
 632
 633                 // Then check if a message gives a converted name in content language
 634                 // which needs extra translation to the target variant.
 635                 if ( $nsVariantText === false ) {
 636                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 637                         if ( $nsConvMsg->exists() ) {
 638                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 639                         }
 640                 }
 641
 642                 if ( $nsVariantText === false ) {
 643                         // No message exists, retrieve it from the target variant's namespace names.
 644                         $langObj = $this->mLangObj->factory( $variant );
 645                         $nsVariantText = $langObj->getFormattedNsText( $index );
 646                 }
 647
 648                 $cache->set( $key, $nsVariantText, 60 );
 649
 650                 return $nsVariantText;
 651         }
 652
 653         /**
 654          * Convert text to different variants of a language. The automatic
 655          * conversion is done in autoConvert(). Here we parse the text
 656          * marked with -{}-, which specifies special conversions of the
 657          * text that can not be accomplished in autoConvert().
 658          *
 659          * Syntax of the markup:
 660          * -{code1:text1;code2:text2;...}-  or
 661          * -{flags|code1:text1;code2:text2;...}-  or
 662          * -{text}- in which case no conversion should take place for text
 663          *
 664          * @warning Glossary state is maintained between calls. Never feed this
 665          *   method input that hasn't properly been escaped as it may result in
 666          *   an XSS in subsequent calls, even if those subsequent calls properly
 667          *   escape things.
 668          * @param string $text Text to be converted, already html escaped.
 669          * @return string Converted text (html)
 670          */
 671         public function convert( $text ) {
 672                 $variant = $this->getPreferredVariant();
 673                 return $this->convertTo( $text, $variant );
 674         }
 675
 676         /**
 677          * Same as convert() except a extra parameter to custom variant.
 678          *
 679          * @param string $text Text to be converted, already html escaped
 680          * @param-taint $text exec_html
 681          * @param string $variant The target variant code
 682          * @return string Converted text
 683          * @return-taint escaped
 684          */
 685         public function convertTo( $text, $variant ) {
 686                 global $wgDisableLangConversion;
 687                 if ( $wgDisableLangConversion ) {
 688                         return $text;
 689                 }
 690                 // Reset converter state for a new converter run.
 691                 $this->mConvRuleTitle = false;
 692                 return $this->recursiveConvertTopLevel( $text, $variant );
 693         }
 694
 695         /**
 696          * Recursively convert text on the outside. Allow to use nested
 697          * markups to custom rules.
 698          *
 699          * @param string $text Text to be converted
 700          * @param string $variant The target variant code
 701          * @param int $depth Depth of recursion
 702          * @return string Converted text
 703          */
 704         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 705                 $startPos = 0;
 706                 $out = '';
 707                 $length = strlen( $text );
 708                 $shouldConvert = !$this->guessVariant( $text, $variant );
 709                 $continue = 1;
 710
 711                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 712                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 713                 // phpcs:ignore Generic.Files.LineLength
 714                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 715                 while ( $startPos < $length && $continue ) {
 716                         $continue = preg_match(
 717                                 // Only match -{ outside of html.
 718                                 "/$noScript|$noStyle|$noHtml|-\{/",
 719                                 $text,
 720                                 $m,
 721                                 PREG_OFFSET_CAPTURE,
 722                                 $startPos
 723                         );
 724
 725                         if ( !$continue ) {
 726                                 // No more markup, append final segment
 727                                 $fragment = substr( $text, $startPos );
 728                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 729                                 return $out;
 730                         }
 731
 732                         // Offset of the match of the regex pattern.
 733                         $pos = $m[0][1];
 734
 735                         // Append initial segment
 736                         $fragment = substr( $text, $startPos, $pos - $startPos );
 737                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 738                         // -{ marker found, not in attribute
 739                         // Advance position up to -{ marker.
 740                         $startPos = $pos;
 741                         // Do recursive conversion
 742                         // Note: This passes $startPos by reference, and advances it.
 743                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 744                 }
 745                 return $out;
 746         }
 747
 748         /**
 749          * Recursively convert text on the inside.
 750          *
 751          * @param string $text Text to be converted
 752          * @param string $variant The target variant code
 753          * @param int &$startPos
 754          * @param int $depth Depth of recursion
 755          *
 756          * @throws MWException
 757          * @return string Converted text
 758          */
 759         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 760                 // Quick sanity check (no function calls)
 761                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 762                         throw new MWException( __METHOD__ . ': invalid input string' );
 763                 }
 764
 765                 $startPos += 2;
 766                 $inner = '';
 767                 $warningDone = false;
 768                 $length = strlen( $text );
 769
 770                 while ( $startPos < $length ) {
 771                         $m = false;
 772                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 773                         if ( !$m ) {
 774                                 // Unclosed rule
 775                                 break;
 776                         }
 777
 778                         $token = $m[0][0];
 779                         $pos = $m[0][1];
 780
 781                         // Markup found
 782                         // Append initial segment
 783                         $inner .= substr( $text, $startPos, $pos - $startPos );
 784
 785                         // Advance position
 786                         $startPos = $pos;
 787
 788                         switch ( $token ) {
 789                                 case '-{':
 790                                         // Check max depth
 791                                         if ( $depth >= $this->mMaxDepth ) {
 792                                                 $inner .= '-{';
 793                                                 if ( !$warningDone ) {
 794                                                         $inner .= '<span class="error">' .
 795                                                                 wfMessage( 'language-converter-depth-warning' )
 796                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 797                                                                 '</span>';
 798                                                         $warningDone = true;
 799                                                 }
 800                                                 $startPos += 2;
 801                                                 break;
 802                                         }
 803                                         // Recursively parse another rule
 804                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 805                                         break;
 806                                 case '}-':
 807                                         // Apply the rule
 808                                         $startPos += 2;
 809                                         $rule = new ConverterRule( $inner, $this );
 810                                         $rule->parse( $variant );
 811                                         $this->applyManualConv( $rule );
 812                                         return $rule->getDisplay();
 813                                 default:
 814                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 815                         }
 816                 }
 817
 818                 // Unclosed rule
 819                 if ( $startPos < $length ) {
 820                         $inner .= substr( $text, $startPos );
 821                 }
 822                 $startPos = $length;
 823                 return '-{' . $this->autoConvert( $inner, $variant );
 824         }
 825
 826         /**
 827          * If a language supports multiple variants, it is possible that
 828          * non-existing link in one variant actually exists in another variant.
 829          * This function tries to find it. See e.g. LanguageZh.php
 830          * The input parameters may be modified upon return
 831          *
 832          * @param string &$link The name of the link
 833          * @param Title &$nt The title object of the link
 834          * @param bool $ignoreOtherCond To disable other conditions when
 835          *   we need to transclude a template or update a category's link
 836          */
 837         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 838                 # If the article has already existed, there is no need to
 839                 # check it again, otherwise it may cause a fault.
 840                 if ( is_object( $nt ) && $nt->exists() ) {
 841                         return;
 842                 }
 843
 844                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 845                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 846                 $action = $wgRequest->getText( 'action' );
 847                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 848                         $action = 'view';
 849                 }
 850                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 851                 $disableLinkConversion = $wgDisableLangConversion
 852                         || $wgDisableTitleConversion;
 853                 $linkBatch = new LinkBatch();
 854
 855                 $ns = NS_MAIN;
 856
 857                 if ( $disableLinkConversion ||
 858                         ( !$ignoreOtherCond &&
 859                                 ( $isredir == 'no'
 860                                         || $action == 'edit'
 861                                         || $action == 'submit'
 862                                         || $linkconvert == 'no' ) ) ) {
 863                         return;
 864                 }
 865
 866                 if ( is_object( $nt ) ) {
 867                         $ns = $nt->getNamespace();
 868                 }
 869
 870                 $variants = $this->autoConvertToAllVariants( $link );
 871                 if ( !$variants ) { // give up
 872                         return;
 873                 }
 874
 875                 $titles = [];
 876
 877                 foreach ( $variants as $v ) {
 878                         if ( $v != $link ) {
 879                                 $varnt = Title::newFromText( $v, $ns );
 880                                 if ( !is_null( $varnt ) ) {
 881                                         $linkBatch->addObj( $varnt );
 882                                         $titles[] = $varnt;
 883                                 }
 884                         }
 885                 }
 886
 887                 // fetch all variants in single query
 888                 $linkBatch->execute();
 889
 890                 foreach ( $titles as $varnt ) {
 891                         if ( $varnt->getArticleID() > 0 ) {
 892                                 $nt = $varnt;
 893                                 $link = $varnt->getText();
 894                                 break;
 895                         }
 896                 }
 897         }
 898
 899         /**
 900          * Returns language specific hash options.
 901          *
 902          * @return string
 903          */
 904         public function getExtraHashOptions() {
 905                 $variant = $this->getPreferredVariant();
 906
 907                 return '!' . $variant;
 908         }
 909
 910         /**
 911          * Guess if a text is written in a variant. This should be implemented in subclasses.
 912          *
 913          * @param string $text The text to be checked
 914          * @param string $variant Language code of the variant to be checked for
 915          * @return bool True if $text appears to be written in $variant, false if not
 916          *
 917          * @author Nikola Smolenski <smolensk@eunet.rs>
 918          * @since 1.19
 919          */
 920         public function guessVariant( $text, $variant ) {
 921                 return false;
 922         }
 923
 924         /**
 925          * Load default conversion tables.
 926          * This method must be implemented in derived class.
 927          *
 928          * @private
 929          * @throws MWException
 930          */
 931         function loadDefaultTables() {
 932                 $class = static::class;
 933                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 934         }
 935
 936         /**
 937          * Load conversion tables either from the cache or the disk.
 938          * @private
 939          * @param bool $fromCache Load from memcached? Defaults to true.
 940          */
 941         function loadTables( $fromCache = true ) {
 942                 global $wgLanguageConverterCacheType;
 943
 944                 if ( $this->mTablesLoaded ) {
 945                         return;
 946                 }
 947
 948                 $this->mTablesLoaded = true;
 949                 $this->mTables = false;
 950                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 951                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 952                 if ( $fromCache ) {
 953                         $this->mTables = $cache->get( $cacheKey );
 954                 }
 955                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 956                         // not in cache, or we need a fresh reload.
 957                         // We will first load the default tables
 958                         // then update them using things in MediaWiki:Conversiontable/*
 959                         $this->loadDefaultTables();
 960                         foreach ( $this->mVariants as $var ) {
 961                                 $cached = $this->parseCachedTable( $var );
 962                                 $this->mTables[$var]->mergeArray( $cached );
 963                         }
 964
 965                         $this->postLoadTables();
 966                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 967
 968                         $cache->set( $cacheKey, $this->mTables, 43200 );
 969                 }
 970         }
 971
 972         /**
 973          * Hook for post processing after conversion tables are loaded.
 974          */
 975         function postLoadTables() {
 976         }
 977
 978         /**
 979          * Reload the conversion tables.
 980          *
 981          * Also used by test suites which need to reset the converter state.
 982          *
 983          * @private
 984          */
 985         private function reloadTables() {
 986                 if ( $this->mTables ) {
 987                         unset( $this->mTables );
 988                 }
 989
 990                 $this->mTablesLoaded = false;
 991                 $this->loadTables( false );
 992         }
 993
 994         /**
 995          * Parse the conversion table stored in the cache.
 996          *
 997          * The tables should be in blocks of the following form:
 998          *              -{
 999          *                      word => word ;
1000          *                      word => word ;
1001          *                      ...
1002          *              }-
1003          *
1004          * To make the tables more manageable, subpages are allowed
1005          * and will be parsed recursively if $recursive == true.
1006          *
1007          * @param string $code Language code
1008          * @param string $subpage Subpage name
1009          * @param bool $recursive Parse subpages recursively? Defaults to true.
1010          *
1011          * @return array
1012          */
1013         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
1014                 static $parsed = [];
1015
1016                 $key = 'Conversiontable/' . $code;
1017                 if ( $subpage ) {
1018                         $key .= '/' . $subpage;
1019                 }
1020                 if ( array_key_exists( $key, $parsed ) ) {
1021                         return [];
1022                 }
1023
1024                 $parsed[$key] = true;
1025
1026                 if ( $subpage === '' ) {
1027                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
1028                 } else {
1029                         $txt = false;
1030                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
1031                         if ( $title && $title->exists() ) {
1032                                 $revision = Revision::newFromTitle( $title );
1033                                 if ( $revision ) {
1034                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
1035                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
1036                                         }
1037
1038                                         // @todo in the future, use a specialized content model, perhaps based on json!
1039                                 }
1040                         }
1041                 }
1042
1043                 # Nothing to parse if there's no text
1044                 if ( $txt === false || $txt === null || $txt === '' ) {
1045                         return [];
1046                 }
1047
1048                 // get all subpage links of the form
1049                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1050                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1051                         ':Conversiontable';
1052                 $subs = StringUtils::explode( '[[', $txt );
1053                 $sublinks = [];
1054                 foreach ( $subs as $sub ) {
1055                         $link = explode( ']]', $sub, 2 );
1056                         if ( count( $link ) != 2 ) {
1057                                 continue;
1058                         }
1059                         $b = explode( '|', $link[0], 2 );
1060                         $b = explode( '/', trim( $b[0] ), 3 );
1061                         if ( count( $b ) == 3 ) {
1062                                 $sublink = $b[2];
1063                         } else {
1064                                 $sublink = '';
1065                         }
1066
1067                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1068                                 $sublinks[] = $sublink;
1069                         }
1070                 }
1071
1072                 // parse the mappings in this page
1073                 $blocks = StringUtils::explode( '-{', $txt );
1074                 $ret = [];
1075                 $first = true;
1076                 foreach ( $blocks as $block ) {
1077                         if ( $first ) {
1078                                 // Skip the part before the first -{
1079                                 $first = false;
1080                                 continue;
1081                         }
1082                         $mappings = explode( '}-', $block, 2 )[0];
1083                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1084                         $table = StringUtils::explode( ';', $stripped );
1085                         foreach ( $table as $t ) {
1086                                 $m = explode( '=>', $t, 3 );
1087                                 if ( count( $m ) != 2 ) {
1088                                         continue;
1089                                 }
1090                                 // trim any trailling comments starting with '//'
1091                                 $tt = explode( '//', $m[1], 2 );
1092                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1093                         }
1094                 }
1095
1096                 // recursively parse the subpages
1097                 if ( $recursive ) {
1098                         foreach ( $sublinks as $link ) {
1099                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1100                                 $ret = $s + $ret;
1101                         }
1102                 }
1103
1104                 if ( $this->mUcfirst ) {
1105                         foreach ( $ret as $k => $v ) {
1106                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1107                         }
1108                 }
1109                 return $ret;
1110         }
1111
1112         /**
1113          * Enclose a string with the "no conversion" tag. This is used by
1114          * various functions in the Parser.
1115          *
1116          * @param string $text Text to be tagged for no conversion
1117          * @param bool $noParse Unused
1118          * @return string The tagged text
1119          */
1120         public function markNoConversion( $text, $noParse = false ) {
1121                 # don't mark if already marked
1122                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1123                         return $text;
1124                 }
1125
1126                 $ret = "-{R|$text}-";
1127                 return $ret;
1128         }
1129
1130         /**
1131          * Convert the sorting key for category links. This should make different
1132          * keys that are variants of each other map to the same key.
1133          *
1134          * @param string $key
1135          *
1136          * @return string
1137          */
1138         function convertCategoryKey( $key ) {
1139                 return $key;
1140         }
1141
1142         /**
1143          * Refresh the cache of conversion tables when
1144          * MediaWiki:Conversiontable* is updated.
1145          *
1146          * @param Title $titleobj The Title of the page being updated
1147          */
1148         public function updateConversionTable( Title $titleobj ) {
1149                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1150                         $title = $titleobj->getDBkey();
1151                         $t = explode( '/', $title, 3 );
1152                         $c = count( $t );
1153                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1154                                 if ( $this->validateVariant( $t[1] ) ) {
1155                                         $this->reloadTables();
1156                                 }
1157                         }
1158                 }
1159         }
1160
1161         /**
1162          * Get the cached separator pattern for ConverterRule::parseRules()
1163          * @return string
1164          */
1165         function getVarSeparatorPattern() {
1166                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1167                         // varsep_pattern for preg_split:
1168                         // text should be splited by ";" only if a valid variant
1169                         // name exist after the markup, for example:
1170                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1171                         //  <span style="font-size:120%;">yyy</span>;}-
1172                         // we should split it as:
1173                         //  [
1174                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1175                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1176                         //    [2] => ''
1177                         //  ]
1178                         $pat = '/;\s*(?=';
1179                         foreach ( $this->mVariants as $variant ) {
1180                                 // zh-hans:xxx;zh-hant:yyy
1181                                 $pat .= $variant . '\s*:|';
1182                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1183                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1184                         }
1185                         $pat .= '\s*$)/';
1186                         $this->mVarSeparatorPattern = $pat;
1187                 }
1188                 return $this->mVarSeparatorPattern;
1189         }
1190 }